def test_contains_verb_noun_pair_5(self):
caption = "a man is sitting on a chair."
nouns = {"man"}
verbs = {"sit"}
pos_tagged_caption = self.nlp_pipeline(caption).sentences[0]
expected_pattern = (True, True, True)
self.assertEqual(
expected_pattern, contains_verb_noun_pair(pos_tagged_caption, nouns, verbs)
)
def count_verb_noun_pairs(nouns_file, verbs_file, preprocessed_data_folder):
with open(nouns_file, "r") as json_file:
nouns = json.load(json_file)
with open(verbs_file, "r") as json_file:
verbs = json.load(json_file)
with open(
os.path.join(preprocessed_data_folder, POS_TAGGED_CAPTIONS_FILENAME), "rb"
) as pickle_file:
captions = pickle.load(pickle_file)
first_noun = nouns[0]
first_verb = verbs[0]
print("Looking for pairs: {} - {}".format(verbs, nouns))
data = {}
data[NOUNS] = nouns
data[VERBS] = verbs
occurrence_data = {}
for coco_id, tagged_caption in tqdm(captions.items()):
occurrence_data[coco_id] = {}
occurrence_data[coco_id][PAIR_OCCURENCES] = 0
occurrence_data[coco_id][VERB_OCCURRENCES] = 0
occurrence_data[coco_id][NOUN_OCCURRENCES] = 0
occurrence_data[coco_id][DATA_COCO_SPLIT] = tagged_caption[DATA_COCO_SPLIT]
for caption in tagged_caption["pos_tagged_captions"]:
noun_is_present, verb_is_present, combination_is_present = contains_verb_noun_pair(
caption, nouns, verbs
)
if combination_is_present:
print(" ".join([token.text for token in caption.tokens]))
occurrence_data[coco_id][PAIR_OCCURENCES] += 1
if verb_is_present:
occurrence_data[coco_id][VERB_OCCURRENCES] += 1
if noun_is_present:
occurrence_data[coco_id][NOUN_OCCURRENCES] += 1
data[OCCURRENCE_DATA] = occurrence_data
data_path = "{}_{}.json".format(first_verb, first_noun)
print("\nSaving results to {}".format(data_path))
with open(data_path, "w") as json_file:
json.dump(data, json_file)
for n in range(1, 6):
noun_occurences = len(
[d for d in occurrence_data.values() if d[NOUN_OCCURRENCES] >= n]
)
verb_occurences = len(
[d for d in occurrence_data.values() if d[VERB_OCCURRENCES] >= n]
)
pair_occurences = len(
[d for d in occurrence_data.values() if d[PAIR_OCCURENCES] >= n]
)
print(
"\nFound {}\timages where the noun occurs at least {} time(s).".format(
noun_occurences, n
)
)
print(
"Found {}\timages where the verb occurs at least {} time(s).".format(
verb_occurences, n
)
)
print(
"Found {}\timages where the pair occurs at least {} time(s).".format(
pair_occurences, n
)
)